New Model for Simulating Impact of
Negative Bias Temperature Instability (NBTI)
in CMOS Circuits
by
Sudheer Padala
A Thesis Presented in Partial Fulfillment
of the Requirements for the Degree
Master of Science
Approved November 2014 by the
Graduate Supervisory Committee:
Hugh Barnaby, Chair
Bertan Bakkaloglu
Jennifer Kitchen
ARIZONA STATE UNIVERSITY
December 2014
i
ABSTRACT
Negative Bias Temperature Instability (NBTI) is commonly seen in p-channel
transistors under negative gate voltages at an elevated temperature. The interface
traps, oxide traps and NBTI mechanisms are discussed and their effect on circuit
degradation and results are discussed. This thesis focuses on developing a model for
simulating impact of NBTI effects at circuit level. The model mimics the effects of
degradation caused by the defects.
The NBTI model developed in this work is validated and sanity checked by using the
simulation data from silvaco and gives excellent results. Furthermore the
susceptibility of CMOS circuits such as the CMOS inverter, and a ring oscillator to
NBTI is investigated. The results show that the oscillation frequency of a ring
oscillator decreases and the SET pulse broadens with the NBTI.
iii
ACKNOWLEDGMENTS
I would like to thank my advisor Dr. Hugh Barnaby for giving me an opportunity to
be part of his research group. His wide knowledge on the topic and his extensive
research has been of great value for me. His understanding, encouraging and personal
guidance have provided a good basis for the present thesis. I want to express my
deepest gratitude to the Ridgetop Group for giving me the opportunity to work with
them as an Intern. It was one of the best learning experiences one could expect to get
as a student.
I am deeply grateful to my thesis committee members, Bertan Bakkaloglu and
Jennifer Kitchen for their comments and valuable time.
I am thankful to Mehdi Saremi with whom I had very good technical discussions. I
also want to thank some of my friends - Raviraj, Harsha, Raghuvir, Yashwanth,
Sravan, Akhilesh, Abhishek, my roommates and many others who helped me and
constantly supported me during my hard times at ASU.
Finally I am indebted to my family for their constant encouragement and belief in me.
I am very proud of them and I constantly work hard to achieve my dreams and to
make them proud.
iv
TABLE OF CONTENTS
Page
LIST OF TABLES …. ....................................................................................................vii
LIST OF FIGURES ……………. ................................................................................. viii
LIST OF ABBREVIATIONS ............................................................................................ x
CHAPTER
1 INTRODUCTION ................ ..................................................................................... 1
2 OVERVIEW OF THE DEFECTS ............................................................................. 2
2.1 MOSFET Analysis ................................................................................... 2
2.1.1 The Basic MOSFET Equations ...................................................... 2
2.1.2 MOSFET Region of Operations ..................................................... 4
2.1.3 Interface and Oxide Charges .......................................................... 5
2.1.4 NBTI Related Models ..................................................................... 5
2.2 The Silicon/Silicon Dioxide Interface ..................................................... 6
2.2.1 The Silicon Dangling Bonds ........................................................... 7
2.2.2 Oxide Traps ...................................................................................... 8
2.2.3 Oxide Charges/Interface Traps ........................................................ 9
3 NEGATIVE BIAS TEMPERATURE INSTABILITY .......................................... 10
3.1 Physical Mecahanism of NBTI .............................................................. 10
3.1.1 Reaction Diffusion Model............................................................. 11
3.1.2 Hydrogen Model ............................................................................. 11
v
CHAPTER Page
3.1.3 Experimental Setup for NBTI .......................................................... 12
3.2 How p-MOSFET are different from n-MOSFET.......................................... 13
3.2.1 Occupancy of Interface Traps ........................................................ 13
3.2.2 Breaking of Si-H Bonds ................................................................. 15
3.2.3 Surface potential-gate voltage behavior ........................................ 16
4 MODELING OF NBTI ............................................................................................ 18
4.1 Modelling the defects in a PMOS transistor.......................................... 18
4.2 Modelling of defects in NMOS transistor ............................................. 23
4.3 Validation and Results ........................................................................... 23
4.3.1 IV characteristics ............................................................................ 25
4.3.2 Threshold Voltage Shift ................................................................. 27
4.4 Modeling Defect Generators .................................................................. 27
4.4.1 Interface Traps (Dit) Generator ..................................................... 28
4.4.2 Oxide Traps (Not) Generator ......................................................... 30
4.5 Static NBTI ............................................................................................. 32
4.6 Dynamic NBTI ....................................................................................... 33
5 NBTI EFFECTS IN A CMOS CIRCUITS ................................................................ 34
5.1 CMOS Inverter ....................................................................................... 34
5.1.1 NBTI Effects in CMOS Inverter .................................................. 36
5.1.2 Inverter with the degradation model ............................................ 37
5.2 Ring Oscillator ....................................................................................... 39
vi
CHAPTER Page
6 CONCLUSION........................................................................................................... 44
REFRENCES .................................................................................................................... 45
APPENDIX
A TCAD CODE ................................................................................................... 53
vii
LIST OF TABLES
Table Page
1. Oxide Charges/Interface Traps ................................................................................ 9
2. Bias Stress Temperature in Inverter ...................................................................... 36
3. Component parameters of Inverter ........................................................................ 38
viii
LIST OF FIGURES
Figure Page
1. Basic n-MOSFET Structure ......................................................................... 2
2. Channel Formation in p-Substrate ............................................................... 3
3. Si-SiO2 Interface and Oxide Defect Structure ............................................ 6
4. The Valence Electrons Forms Active Interface Traps at the Surface ........ 7
5. The Interface Traps Are Reduced after Oxidation ...................................... 7
6. The Interface Traps Are More Suppressed after Hydrogen Passivation .... 8
7. Oxide Charges and the Interface Traps........................................................ 9
8. Reaction-Diffusion Model ......................................................................... 11
9. Hydrogen Model ......................................................................................... 12
10. The Setup for NBTI Experiment................................................................ 12
11. BTI in PMOS and NMOS Transistors ....................................................... 13
12. Band Diagram Showing the Polarities of Interface States ........................ 14
13. Interface Trap Creation Due to Hydrogen [10] ......................................... 16
14. Simulation Setup for PMOS with the Degradation Model ....................... 19
15. Degradation Model Schematic for a PMOS Transistor ............................ 22
16. p- Channel MOSFET Silvaco Structure ................................................... 24
17. Validation Model Block Diagram ............................................................. 24
18. I-V Curves of PMOS for Not = 0 and Dit = 0. ......................................... 25
19. I-V Curves of PMOS with Not=1.2 x1012
cm-2
and Dit=1x1012
cm-2
...... 25
20. I-V Curves of PMOS with Not = 0 and Dit = 1 x 1012
cm-2
..................... 26
21. I-V Curves of PMOS with Not = 1.2 x 1012
cm-2
and Dit = 0 .................. 26
ix
Figure Page
22. Threshold Voltage Before and After the Degradation Model ................... 27
23. Threshold Voltage Shift- Defect Model and Vt only…...……………. 27
24. Ideal Response of Interface Traps During Stress and Relax ..................... 28
25. Nit Defect Generator Schematic ................................................................ 29
26. Interface Trap Voltage with Time and Gate Bias ...................................... 30
27. Ideal Response of Oxide Traps During Stress and Relax ......................... 30
28. Not Defect Generator Schematic ............................................................... 31
29. Oxide Trap Voltage with Time and Gate Bias .......................................... 32
30. Static NBTI ................................................................................................. 33
31. Dynamic NBTI ........................................................................................... 33
32. CMOS Inverter ........................................................................................... 34
33. The CMOS Inverter Voltage Transfer Characteristics .............................. 35
34. Inverter Propagation Delay ....................................................................... 35
35. CMOS Inverter Degradation ...................................................................... 37
36. CMOS Inverter with and Without Degradation Models .......................... 38
37. CMOS Inverter Output with and Without Stress ..................................... 39
38. An n-Stage Ring Oscillator ........................................................................ 39
39. Ring Oscillator Schematic with 25 Inverters ............................................. 41
40. Ring Oscillator Output Waveform with and Without Stress .................... 41
41. Frequency Response Without Stress.......................................................... 42
42. Frequency Response with Stress(Defects Model & Vt Model) ............... 42
43. Modeling SET Pulse Broadening .............................................................. 43
44. SET Pulse Broadening Without and with Stress ....................................... 43
x
LIST OF ABBREVIATIONS
MOSFET Metal-Oxide-Semiconductor Field-Effect Transistor
nMOSFET n-channel MOSFET
pMOSFET p-channel MOSFET
NBTI Negative Bias Temperature Instability
PBTI Positive Bias Temperature Instability
BTI Bias Temperature Instability
CMOS Complementary Metal–Oxide–Semiconductor
PMOS p-Channel Metal-Oxide-Semiconductor
NMOS n-Channel Metal-Oxide-Semiconductor
Dit Density of Interface Traps
Not Density of Oxide Traps
DOS Density of States
BTS Bias Temperature Stress
TCAD Technology Computer Aided Design
SET Single Event Transient
1
CHAPTER 1
INTRODUCTION
Negative-bias temperature instability (NBTI) is a main reliability issue in
MOSFETs. The NBTI effect is more seen in a p-channel MOSFET when stressed
with negative gate voltages at elevated temperatures. The NBTI causes the absolute
value of threshold voltage to increase and consequent decrease in drain current and
decrease of other important parameters like trans-conductance, channel mobility of a
MOSFET. The same effects can be seen in n-channel MOSFET’s too but the
degradation is very less compared to that of a p-channel MOSFET. Therefore it is of
immediate concern in p-channel MOS devices, since they almost always operate with
negative gate-to-source voltage like in an inverter and ring oscillator.
There are two main types of traps which contribute to NBTI:
Interface traps: These traps cannot be recovered over a reasonable time of
operation. In the case of NBTI, it is believed that the Si–H bonds located at the
silicon-silicon dioxide interface are broken when interacted with the protons
coming from substrate [42]. The dangling bonds contribute to the threshold
voltage degradation.
Oxide traps: When gate stress is applied, some holes are pulled into the oxide
from channel of PMOS. These traps can be emptied when the stress voltage is
removed. The threshold voltage degradation can be recovered over time.
A new model to simulate the circuit level degradation subjected to negative bias
temperature instability (NBTI) has been developed in this thesis work. A very basic
2
fundamental ring oscillator circuit is studied to evaluate the impact of NBTI on
CMOS digital circuit performances.
The change in threshold voltage is modeled as a defect potential and is applied at gate.
The defects dependence on time and gate voltage is modeled in the defect potential.
3
CHAPTER 2
OVERVIEW OF THE DEFECTS
Before even discussing about the defects, let us discuss about the basic MOSFET
operation like the channel formation, regions of operation and the threshold voltage.
2.1 MOSFET ANALYSIS
2.1.1 The Basic MOSFET Operation
A metal–oxide–semiconductor field-effect transistor (MOSFET) is basically a
device which controls and modulates the charge by a capacitance between a body
terminal and a gate terminal. The gate of the MOSFET is located on the top of the
body substrate and it is insulated from other regions by a field oxide, such as silicon
dioxide. The body of the MOSFET is generally doped by silicon. The p-channel
MOSFET substrate is doped by n-silicon, whereas the n-channel MOSFET substrate
is doped by p-silicon. The silicon surface and gate terminal acts like two terminals of
the capacitance and the dielectric is silicon dioxide [12].
In nMOSFET, the source and drain regions are "n+" type and the body substrate is "p"
type silicon as shown in Figure 1. In the case of a pMOSFET, the source and drain are
"p+" regions and the body is an "n" region. In pMOSFET, the source and drain
regions are “p” type and body substrate is “n” type silicon.
4
Figure 1: Basic p-MOSFET Structure
Let us see how a channel is formed in an nMOSFET. The same theory applies to
pMOSFETs.
In nMOSFET, the substrate is “p” type where holes are the majority carriers. When a
positive voltage is applied to the gate, it repels the positively charged holes away from
the substrate. A depletion region is formed when the holes move further deep into the
substrate. When the positive voltage is applied at the gate, it attracts electrons from
the source and drain which are “n+” regions and have excess of electrons, into the
channel region. As the number of electrons accumulate near the surface of the Si-
SiO2, an n region is in created, connecting the source and drain regions as shown in
Figure 2. If a voltage is now applied at the drain of the MOSFET, current starts
flowing through this channel. Thus a channel is formed between drain and source and
current flows in a MOSFET [13].
5
Figure 2: Channel Formation in p-Substrate
2.1.2 MOSFET Region of Operations
The MOSFET operates in different regions depending on the gate voltage.
Cutoff, Sub threshold, or weak inversion mode:
When VGS < Vth,
In this region, the transistor is turned off, and there is no current flowing between
drain and source. But there is a weak sub threshold current which is exponentially
dependent on the gate source voltage [12].
ID = ID0e(𝑉𝐺𝑆−𝑉𝑡ℎ)
𝑛𝑉𝑇 (1)
Where ID0 = current at VGS =Vth, the thermal voltage VT = kT/q and the slope factor n
is given by
6
n=1+ 𝐶𝐷
𝐶𝑜𝑥 (2)
Where CD = capacitance of the depletion layer and Cox = capacitance of the oxide
layer. This equation is generally used, but is only an adequate approximation for the
source is connected to the body.
Triode mode or linear region:
When VGS >Vth, VDS < VGS - Vth
In this region the transistor is turned on, and a channel has been formed allowing the
current to flow from drain to source. The MOSFET operates like a resistor, controlled
by the gate voltage relative to both the source and drain voltages.
The drain current is,
Ids = µCox𝑊
𝐿𝑒𝑓𝑓 (Vgs – Vth -
𝑉𝑑𝑠
2 )(1 + λVds) (3)
Where µ is the effective mobility, W is the width of the device, Leff is the effective
gate length and λ is the channel length modulation parameter.
In Saturation region, where
Vgs > Vth and Vds > Vgs – Vth
The transistor is turned on, and a channel has been created, which allows current to
flow between the drain and source [12].
The drain current is,
Ids = 1
2µCox
𝑊
𝐿𝑒𝑓𝑓 (Vgs – Vth)
2 (1 + λVds) (4)
7
The threshold voltage of a MOSFET is given by the equation (5)
Vth = VTo + γ(√2𝜙𝐵 − 𝑉𝐵𝑆 − √2𝜙𝐵 ) + 𝑄𝑖𝑡+𝑄𝑜𝑥
𝐶𝑜𝑥 (5)
Where VTo is the threshold voltage when VBS = 0, Cox is the capacitance per unit area
of the oxide, Qit is the interface charge, Qox is the oxide charge and γ is the body-effect
parameter, defined as
γ= √2𝜀𝑠𝑞𝑜𝑁𝐴
𝐶𝑜𝑥 (6)
Here, εs is the permittivity of the silicon substrate and NA the acceptor doping
concentration. The potential in the neutral p-type region ϕB is evaluated as
ϕB = 𝐾𝑇
𝑞 ln
𝑁𝐴
𝑛𝑖 (7)
During NBTI, the interface defects and oxide traps will be formed which degrades the
performance of the transistor and the threshold voltage is shifted by ΔVth obtained
from (5)
2.1.3 Interface and Oxide Charges
The threshold voltage shift (ΔVth) is contributed mainly because of the
changes in interface and oxide charges (ΔQit and ΔQox) and is given by,
ΔVth = 𝛥𝑄𝑖𝑡+ 𝛥𝑄𝑜𝑥
𝐶𝑜𝑥 (8)
In the context of NBTI interface charges are usually the result of charged interface
defects Dit. The interface charge depends on the Fermi-level EF and the trap
occupancy f and can be calculated as [43]
8
dEEEfEDqQ f
E
Eitit
c
v
, . (9)
Charged oxide traps Dox contribute to the threshold voltage shift depending on their
position in the dielectric. The resulting, effective, Qox can be evaluated as
dxt
xxqQ
ox
t
ot
ox
0 , (9)
where tox is the oxide thickness.
2.1.4 NBTI Related Models
The important parameter which changes because of NBTI is the trans-
conductance (gm). As the drain current changes with the trapped charges, so does the
trans-conductance.
gm = 𝛥𝐼𝑑
𝛥𝑉𝑔 (11)
Devine et al. proposed a trans-conductance shift versus interface trap (Nit), relation as
[45]
Δgm = gmo 𝛼𝑁𝑖𝑡
1+ 𝛼𝑁𝑖𝑡 (12)
Where α is a processing related parameter.
The mobility model proposed by Devine et al. [45] is
µ = 𝜇𝑜
1+ 𝛼𝑁𝑖𝑡 (13)
Therefore the mobility and trans-conductance decreases with increase in the interface
traps. The NBTI also increase the threshold voltage. Decrease in carrier mobility and
9
trans-conductance and increase in absolute value threshold voltage leads to the
reduction of drain current and the performance of the transistor is degraded [20].
2.2 The Silicon/Silicon Dioxide Interface
One of the most important things that led to the enormous success and
continuous improvement of the CMOS technology are the excellent properties of the
thermally grown Si/SiO2 interface.
Figure 3: Si-SiO2 Interface and Oxide Defect Structure [53]
Silicon dioxide (SiO2) is formed when the silicon surface is exposed to oxygen. This
process is called oxidation. Silicon dioxide is a very good electrical insulator and
hence we use it for the electrical isolation of the devices, and also as a component in
MOS transistors. It acts like a dielectric in the MOS capacitance, the two plates being
10
the gate and body electrodes. Figure 3 shows the Si/SiO2 interface and various types
of bonds that exist at the interface and in the oxide.
2.2.1 Silicon Dangling Bonds
The silicon atom has four valence electrons and hence it wants to bond with
four atoms to be stable. In the silicon crystal each silicon atom bonds with four other
silicon atoms but at the surface there are no silicon atoms to bond with and hence
traps are formed as shown in Figure 4. The density of these interface states, Dit is
approximately 1014
cm-2
eV-1
. When the silicon surface is oxidized, some of the
silicon atoms at the surface are bonded with the oxygen atoms as shown in Figure 5.
The density of interface traps then reduces a bit and is approximately Dit = 1012
cm-
2eV
-1 [46].
Figure 4: The Valence Electrons Forms Active Interface Traps at the Surface.
11
Figure 5: The Interface Traps are Reduced After Oxidation.
Figure 6: The Interface Traps are More Suppressed After Hydrogen Passivation.
There can still be a lot of defects present which causes the circuit degradation. So it is
very important that the defects need to be reduces and passivated at the interface in a
transistor. The interface traps causes degradation of transistor parameters like the
threshold voltage, the trans-conductance, the on-current, and the carrier mobility at
device level and increases the delay time at circuit level. These interface traps can be
reduced more by annealing the surface with hydrogen gas. The remaining silicon
dangling bonds react with hydrogen forming Si-H bonds and thus the defects are
reduced to more extent. After the hydrogen passivation the density of interface defects
12
can be reduced to around Dit = 1010
cm-2
eV-1
[46]. These Si-H bonds can break at high
temperatures or with the negative gate stress, forming interface traps [22] as discussed
in the further sections.
2.2.2 Oxide Traps
Oxygen vacancies are the major hole trapping mechanisms. Because of the
manufacturing defects, oxygen vacancies are formed in the oxide which traps the
positively charged holes. When the Si-H bonds break with the gate stress or at high
temperatures, the hydrogen atoms diffuse into the oxide and holes can be trapped [22].
When the gate stress is applied, there will be an electric field in vertical direction
which causes the holes to be trapped. The threshold voltage is degraded because of the
charge accumulated by these holes in the oxide. When the stress is removed, the oxide
traps will be annealed and the threshold voltage can be annealed up to some extent.
The mechanisms for the formation of interface traps and oxide traps are discussed in
the further sections.
2.2.3 Oxide Charges/Interface Traps
Figure 7: Oxide Charges and the Interface Traps [53]
13
Table 1 gives more details like the causes and effects of the oxide charges and
interface traps which were shown in Figure 7.
Table 1: Oxide Charges/Interface Traps [53]
Charge Type Location Cause Effect on Devices
(1)
Dit(cm-2eV
-1),
Nit, Qit
Interface
Trap
Charge
SiO2/Si
Interface
Dangling
Bonds
Junction Leakage
Current Noise,
Threshold Voltage
Shift, Sub Threshold
Slope
(2)
Nf, Qf
cm-2
, C/cm2
Fixed
Charge
Close to
SiO2/Si
Interface
Si+
Threshold Voltage
Shift
(3)
Not, Qot
Oxide
Trapped
Charge
In SiO2 Trapped
Electrons
and
Holes
Threshold Voltage
Shift
Nm, Qm Mobile
Charge
In SiO2 Na, K, Li Threshold Voltage
Shift
In this work, only interface trap charges and oxide trap charges are considered and
their time dependence and gate voltage dependence are modeled to simulate the
impact of NBTI at circuit level.
14
CHAPTER 3
NEGATIVE BIAS TEMPERATURE INSTABILITY (NBTI)
Bias temperature instability (BTI) is a degradation phenomenon affecting
mainly MOS field effect transistors. The highest impact is observed in p-MOSFETs.
Either Negative voltages or high temperatures can cause NBTI, but a faster and
stronger effect is produced with negative stress and high temperature together. The
stress conditions for this negative bias temperature instability (NBTI) typically lie
below 6MV/cm for the gate oxide electric field and temperatures ranging 100-250 ˚C.
A very interesting thing about NBTI is its capability to anneal to a certain extent when
the stress is removed [18].
The important transistor parameters which degrade because of NBTI are:
Trans-conductance (gm) decreases
Linear drain current and saturation current (Idsat) decreases
Channel mobility (µeff) decreases
Sub-threshold slope (S) decreases
Absolute value of the threshold voltage (Vth) increases
At the circuit level the switching speeds are reduced and the delay times are increased
as the charging time for load capacitances or the parasitic capacitances are increased.
The NBTI is first reported many years ago [49], but gained much attention in recent
years [3, 11, 18, 48] due to modern semiconductor technologies.
15
The following aspects are the main factors that lead to the increasing NBTI [20]:
higher oxide electric fields due to oxide scaling
higher temperatures due to higher power dissipation
high performance IC’s during routine operation
3.1 Physical Mechanisms of NBTI
The negative bias temperature instability (NBTI) effect has been reported many
years ago by several groups [49, 50, 51], there is still a lot to understand the exact
mechanisms behind the NBTI degradation. The general theory says that when
MOSFETs are stressed with a gate voltage at an elevated temperature, charge builds
up because of the interface traps and the oxide traps at the interface and oxide
respectively. This charge leads to degradation of the transistor performance [20].
PMOS transistors are more susceptible to the degradation as they always operate with
negative gate source voltage.
3.1.1 Reaction Diffusion Model
The reaction-diffusion (R-D) model is been proposed by several groups [7, 8,
9] states that when a p-MOSFET is in strong inversion, the holes in the inversion
layer reacts with Si-H bonds at the Si-SiO2 interface and weakens the bonds. At high
temperatures, these Hydrogen bonds dissociate forming active interface traps (Nit).
Initially the generation rate of interface traps depends on the dissociation of the Si-H
bonds. This process is reaction. Later the generation of interface traps depends on the
16
diffusion of hydrogen atoms. This is the diffusion step [22]. Figure 8 shows the
reaction diffusion model for NBTI.
Figure 8: Reaction-Diffusion Model
𝑑𝑁𝑖𝑡
𝑑𝑡 = Kf(NSi-H – Nit) - KrNitNH(x=0) (14)
Where Kf, Kr, NSi-H and NH(x=0) are bond-breaking rate, hydrogen annealing rate, Si-
H density before stress and hydrogen density at the interface respectively [22].
3.1.2 Hydrogen Model
The hydrogen atoms from the Si substrate migrate to the Si/SiO2 interface.
These hydrogen atoms are produced when the Phosphorous-Hydrogen bonds (n-
17
substrate) or the Boron-Hydrogen bonds (p-substrate) break. The hydrogen atoms
when migrating towards are the interface traps a hole and becomes H+. The positive
hydrogen atoms depassivate the Si-H bonds at the Si/SiO2 interface forming H2 and an
interface trap [42] as shown in Figure 9. In the process, some H+
can go into the oxide,
hence forming oxide traps (Not).
Si3≡Si-H + H+ = Si3≡Si• + H2 (15)
Si• is a dangling bond.
Figure 9: Hydrogen Model [18]
When the stress is removed the H2 can passivate the dangling bonds and the
degradation is reduced.
3.1.3 Experimental Setup for NBTI
The gate is biased with a negative voltage, while the drain, source, and bulk
are grounded.
18
Figure 10: The Setup for NBTI Experiment.
The set-up for the negative bias temperature instability (NBTI) at a device level is
shown in Figure 10. The bulk of the transistor is grounded. The source and drain
contacts are also grounded. A negative gate voltage stress is applied to the gate at
elevated temperatures, typically ranging between 100 and 250 ˚C, for a certain period
of time. To measure the degradation, the threshold voltage can be measured by
plotting the drain current with respect to source gate voltage.
3.2 How p-MOSFET are different from n-MOSFET
NBTI is observed both in PMOS and NMOS transistors; however PMOS shows more
degradation in their performance compared to the NMOS. Figure 11, presenting data
from Huard et al. [3], shows that the threshold voltage degradation is more in PMOS
transistors with negative gate voltage stress. When a positive gate voltage is applied to
PMOS, the degradation is very less compared to that of negative voltage stress and
also the effects in NMOS transistors for both positive and negative voltage are less
too.
19
Figure 11: BTI in PMOS and NMOS Transistors [3].
In the next sections we will discuss the other possible reasons for the different
degradation performance in PMOS and NMOS transistors.
3.2.1 Occupancy of Interface Traps
A donor trap can be either positive or neutral. A donor trap is positively charged when
empty and neutral when filled. An acceptor trap can be either negative or neutral. An
acceptor trap is neutral when empty and negatively charged when filled. The donor
traps usually lie near the valence band and the acceptor traps usually lie near the
conduction band. Figure 12 shows the polarities of the interface trap charge in both
PMOS and NMOS transistors. The lines are the energy states of interface traps and
the solid ones are filled energy states. The charge state of the interface traps depends
on the Fermi level. All the energy states below the Fermi level are filled. The charge
polarity of the interface traps at the Si/SiO2 interface in inversion is important as the
threshold voltage depends on this charge [10].
20
Figure12: Band Diagram Showing the Polarities of Interface States. (a) n- Substrate at
Flat Band and in Inversion. (b) p-Substrate at Flat Band and in Inversion. [10]
In the case of a PMOS, it’s an n substrate. At flat band the acceptor traps are filled
and hence it’s negatively charged. In inversion, the Fermi level is below the intrinsic
energy level and hence the donor traps are empty which means it is positively charged
as shown in Figure 12(a). Hence for an n-substrate in inversion, the charge because of
interface traps is positive [10].
QDit = + Qit (16)
In the case of an NMOS the situation is different. Figure 12(b) shows the energy
diagram of an NMOS in flat band and in inversion. At the flat band, the donor traps
are empty and hence it is positively charged but in inversion, the Fermi level is above
the intrinsic energy level. So the acceptor traps are filled and hence negatively
charged. Hence in inversion the net charge because of interface traps for p-substrate is
negative [10].
21
QDit = – Qit (17)
The Oxide traps are mainly because of the holes getting trapped in the oxide. In both
NMOS and PMOS the mechanism is the same that is the hole gets trapped in oxide
and hence it is positively charged in both cases. Therefore the charge contribution
because of oxide traps is almost same in both PMOS and NMOS and it is positive.
QNot = + Qot (18)
So when we add the net charge contributions caused by interface traps and oxide traps
for PMOS and NMOS, we get
Qtotal = QDit + QNot (19)
For PMOS,
Qtotal = Qit + Qot (20)
For NMOS,
Qtotal = - Qit + Qot (21)
In PMOS, the interface traps and the oxide traps contribute to the threshold voltage
shift whereas in NMOS the interface traps reduces the shift caused by the oxide traps
and the effect of NBTI is therefore less seen in NMOS compared to PMOS devices.
3.2.2 Breaking of Si-H Bonds
Tsetseris et al. [42] proposed a model explaining the different degradation
behavior in p-MOSFETs and n-MOSFETs. The positively charged hydrogen or
22
proton, H+
is responsible for the breaking of Si-H bonds at the interface. The hydrogen
is originated from the phosphorous-hydrogen bonds in the n-substrate in the case of a
p-MOSFET. The hydrogen atom then picks up a hole to form H+ and then reacts with
the Si-H bond at the interface to form H2 and leaved behind a silicon dangling bond
which is an interface trap and the H2 diffuses into the oxide as shown in Figure 13.
Figure 13: Interface Trap Creation Due to Hydrogen [10]
When the stress is removed, the H2 can diffuse back to the interface and hence
passivate the silicon dangling bond.
The activation energy for dissociation of P-H bonds is usually high. The bonds can be
broken only in the depletion region. During negative bias stress a p-MOSFET is in
strong inversion and a depletion region is formed in the substrate. Hydrogen atoms
can break and cause degradation at the interface. But when there is a positive stress
depletion region is not formed. Therefore the protons are not available from the P-H
bonds, and the interface traps are not formed.
23
In the n-MOSFET the substrate is doped with boron and it is more difficult to break
the boron-hydrogen bonds as their binding energy is much higher.
3.2.3 Surface potential-gate voltage behavior
The different degradation behavior of p- and n-channel MOSFETs can be
because the gate voltages can be different for a given oxide electric field [18].
The Fermi-level of a p-type poly gate lies approximately at the valence band edge EF
= EV while the Fermi-level of an n-type poly gate is located at the conduction band
edge energy EF = EC. For NBT stress the p-type poly gate of the p-channel device is
depleted while the n-type poly gate of the n-channel device is driven into
accumulation.
Oxide and interface charges generated by NBTI degradation further shift the relevant
gate voltages.
This asymmetry in n-type and p-type gate contacts, the oxide electric fields are not the
same for a given gate voltage and at a given oxide electric field the gate voltages are
different. This effect has to be accounted while comparing between p-channel
MOSFET’s and n-channel MOSFET’s [20].
24
CHAPTER 4
MODELING OF NBTI
As we have discussed in previous sections, charge build up with the defects
and which degrades the threshold voltage of the transistor. A model has been
developed for simulating impact of Negative Bias Temperature Instability (NBTI) at
circuit level.
4.1 Modelling the defects in a PMOS transistor
If ΔNot and ΔDit are able to be measured experimentally as a function of bias
stress and time, their effects be simulated in Cadence at the circuit level
The approach is to utilize external voltages (ΔV) in series with the transistor gate
stimulus that mimic effects of time and bias dependent defects.
The degradation model is developed using the drain current equation in the sub
threshold region. The sub threshold model works for strong inversion too with slight
modifications.
t
pretSG
preDm
VVI
,
, exp (22)
Step 1
Construct
transistor
degradation
functions
Step 2
Model Transistor
BTI response
(Analytical,
CADENCE, TCAD)
ΔNot , ΔDit
25
t
psttSG
pstDm
VVI
,
, exp (23)
where pretpstt VVV ,, (24)
t
pretSG
pstDm
VVVI
,
, exp (25)
t
pretSG
pstDm
VVI
,
,
*exp (26)
where VVV GG * (27)
Figure 14: Simulation Setup for PMOS with the Degradation Model
Figure 14 shows the simulation setup for PMOS transistor. ΔV is the defect potential.
From (19), we know that for a PMOS transistor the total charge accumulated because
of the interface traps and oxide charge is
ID
ΔV
Defects
Generator
26
Qtotal = Qot + Qit (28)
Therefore
ox
BSit
ox
ot
C
qD
C
qNV
)( (29)
ox
Sit
ox
Bit
ox
ot
C
qD
C
qD
C
qNV
(30)
From (43) BGpstt
oxdm
ox VVCC
C
2,S
where
1 ox
it
C
qD (31)
and B 2S is small
Gpstt
oxdm
it
ox
Bit
ox
ot VVCC
qD
C
qD
C
qNV
,
(32)
Now
pret
ox
Bit
ox
ot
pstt VC
qD
C
qNV ,,
(33)
So
Gpret
ox
Bit
ox
ot
oxdm
it
ox
Bit
ox
ot VVC
qD
C
qN
CC
qD
C
qD
C
qNV ,
(34)
27
The derivation for surface potential ( S )
SSFBG VV (35)
if
BBS 2 and 2
22 BBFBG VV (36)
B
BBFBG VV
2122
(37)
B
BBFBG VV
22
1122 (38)
B
BBFBG VV
22
122
(39)
B
thG VV
22
1 (40)
SB
ox
dm
SBthGC
CVV 22 (41)
where,
𝐶𝑑𝑚 = √εsqNa
4ϕB (42)
28
BGth
oxdm
ox
B
ox
dm
Gth
S VVCC
C
C
C
VV
22
(43)
We know that
1
ox
it
oxdmit
oxdmit
itoxdm
it
oxdm
it
mC
qD
CCqD
CCqD
qDCC
qD
CC
qD
(44)
If, oxdmit CCqD , so using Tylor series
2
oxdmitoxdmit
ox
it CCqDCCqDmC
qD (45)
The function 𝑞𝐷𝑖𝑡
𝑚𝐶𝑜𝑥 is a rational function, so it is approximated.
,
ox
Bit
ox
otpretGB
ox
it
ox
Bit
ox
ot
C
qD
C
qNVV
mC
qD
C
qD
C
qNV
(46)
In Strong Inversion,
For VSG ≥ Vt,post
)( ,
ox
Bit
ox
otpretGB
ox
it
ox
Bit
ox
otG
C
qD
C
qNVV
mC
qD
C
qD
C
qNVV
The terms crossed out should be shorted out with a switch.
The shift ΔV in threshold voltage should be fixed to
ox
Bit
ox
ot
C
qD
C
qNV
(47)
29
This captures the fact that in strong inversion the surface potential is “pinned” at ϕB.
Thus the threshold voltage shift shows no more dependence on VG.
For the 0.25u tsmc technology,
Oxide Thickness tox = 5.8nm.
Threshold Voltage VTHo = 0.4308V.
Substrate Doping NCH = 2.35x1017
cm-3
.
Therefore Bulk potential ϕB = 𝑘𝑇
𝑞 ln (
𝑁𝐶𝐻
𝑛𝑖) =0.456V.
Cdm = √εsqNa
4ϕB = 4.8x10
-7 F/cm
2.
Figure 15 shows the degradation model developed in cadence.
Figure 15: Degradation Model Schematic for a PMOS Transistor
30
4.2 Modelling of defects in NMOS transistor
The same approach used for PMOS applies for NMOS transistor too with slight
changes as shown below.
The total charge accumulated because of defects in NMOS from (21) is
Qtotal = Qot - Qit
Therefore the threshold voltage shift is
ox
BSit
ox
ot
C
qD
C
qNV
)( (48)
After substituting for surface potential and solving for ΔV, we get the defect potential
as
,
ox
Bit
ox
ot
pretG
ox
it
ox
Bit
ox
ot
C
qD
C
qNVV
mC
qD
C
qD
C
qNV
(49)
In Strong Inversion,
For VGS ≥ Vt,post
)( ,
ox
Bit
ox
otpretG
ox
it
ox
Bit
ox
otG
C
qD
C
qNVV
mC
qD
C
qD
C
qNVV
The crossed out terms are shorted out with a switch in strong inversion.
The shift ΔV in threshold voltage should be fixed to
ox
Bit
ox
ot
C
qD
C
qNV
(50)
31
4.3 Validation and Results
The PMOS transistor is modelled in Silvaco (Atlas) as shown in Figure 4.3. The
model parameters are taken from the compact models of 0.25u CMOS technology and
then the I-V characteristics of a transistor are compared both in Cadence (0.25µ) and
TCAD. Once both the curves fit each other, the degradation model is developed in
cadence according to the equation (46). The degradation block is then inserted before
the gate of the transistor and the post stress simulations are done. Similarly in TCAD
the post stress simulations are done with the interface and oxide traps and the I-V
characteristics are compared till both fit each other. The block diagram for validating
the degradation model with TCAD is shown in Figure 16. We validate the model by
fitting the IV curves in cadence model and in TCAD simulations. The comparison is
done for four different values of interface and oxide traps. The same process is
repeated for NMOS transistor.
Figure 16: p- Channel MOSFET Silvaco Structure
32
Figure 17: Validation Model Block Diagram
The model parameters for the TCAD structure are taken from the compact model of
tsmc 0.25u technology. Figure 17 shows the block diagram of validating the model.
4.3.1 IV characteristics
The current voltage characteristics are plotted when the gate voltage is swept
from 2.5V to 0.
33
(a) For Not = 0 and Dit = 0
Figure 18: I-V Curves of PMOS for Not = 0 and Dit = 0.
(b) For Not = 1.2 x 1012
cm-2
and Dit = 1 x 1012
cm-2
Figure 19: I-V Curves of PMOS with Not = 1.2 x 1012
cm-2
and Dit = 1 x 1012
cm-2
.
1.00E-14
1.00E-12
1.00E-10
1.00E-08
1.00E-06
1.00E-04
1.00E-02
1.00E+00
0 0.5 1 1.5 2 2.5 3
Ids_cad
Ids_TCAD
1.00E-18
1.00E-16
1.00E-14
1.00E-12
1.00E-10
1.00E-08
1.00E-06
1.00E-04
1.00E-02
1.00E+00
0 0.5 1 1.5 2 2.5 3
Ids_cad
Ids_TCAD
VSG (V)
Curr
ent
I DS
(A
) VSG (V)
Curr
ent
I DS
(A
)
34
(c) For Not = 0 and Dit = 1 x 1012
cm-2
Figure 20: I-V Curves of PMOS with Not = 0 and Dit = 1 x 1012
cm-2
.
(d) For Not = 1.2 x 1012
cm-2
and Dit = 0
Figure 21: I-V Curves of PMOS with Not = 1.2 x 1012
cm-2
and Dit = 0.
1.00E-15
1.00E-13
1.00E-11
1.00E-09
1.00E-07
1.00E-05
1.00E-03
1.00E-01 0 0.5 1 1.5 2 2.5 3
Ids_cad
Ids_TCAD
1.00E-17
1.00E-15
1.00E-13
1.00E-11
1.00E-09
1.00E-07
1.00E-05
1.00E-03
1.00E-01 0 0.5 1 1.5 2 2.5 3
Ids_cad
Ids_TCAD
VSG (V) C
urr
ent
I DS
(A
)
VSG (V)
Curr
ent
I DS
(A
)
35
As we see from the above curves, the I-V characteristics of the PMOS closely fit each
other in cadence and TCAD. The comparison is made first without any defects and
then with defects for four different cases. In all the cases the degradation model
mimics the defects very well.
4.3.2 Threshold Voltage Shift
Figure 22: Threshold Voltage Before and After the Degradation Model
Figure 23: Threshold Voltage Shift- Defect Model and Vt Only
1.00E-18
1.00E-15
1.00E-12
1.00E-09
1.00E-06
1.00E-03
1.00E+00
0 0.5 1 1.5 2 2.5 3
Ids_pre
Ids_postΔVth
1.00E-18
1.00E-15
1.00E-12
1.00E-09
1.00E-06
1.00E-03
1.00E+00
0 0.5 1 1.5 2 2.5 3
Id_0
Id_Vt only
Id_(Not & Dit)ΔVth
VSG (V)
Curr
ent
I DS
(A
)
VSG (V)
Curr
ent
I DS
(A
)
36
The Current-Voltage curves are generated for a PMOS with and without the
degradation model. We see a threshold voltage shift in the Figure 22, which is caused
by the defects.
Figure 23 shows the current-voltage plots with and without gate bias dependence of
the defects. In the Vt only plot, the change in threshold voltage is given by the
equation (47). Whereas in the other current plot the defects are dependent on gate
voltage in the sub threshold region and the threshold voltage difference is given by the
equation (46).
4.4 Modeling Defect Generators
4.4.1 Interface Traps (Dit) Generator
For interface traps the response profile increases with stress, but remains virtually
constant once the stress is removed. Figure 24 shows the ideal response of interface
traps during stress and when the stress is removed.
Figure 24: Ideal Response of Interface Traps during Stress and Relax
37
This may also be captured by the transient equation but instead with a very large delay
time constant.
VDit(t) = Vpb[(1-exp(- 𝑡
𝜏1))u(t) – (1 – exp( -
𝑡−𝑇
𝜏2))u(t-T)] (51)
where
VDit(t) = 𝑞𝐷𝑖𝑡(𝑡)
𝐶𝑜𝑥 (52)
The equation (51) can be implemented in cadence as shown in Figure 25.
Figure 25: Nit Defect Generator Schematic
The maximum value of Dit is taken to be 1x1012
.
Therefore VDit = 𝑞𝐷𝑖𝑡
𝐶𝑜𝑥 = 268mV.
38
During Stress, the switch is closed and the time constant τ1 = (R1||R2)C.
During Relax, the switch is open and the time constant τ2 = (R2)C.
We want the delay time constant to be very large, so R2 is a big value.
So one can select values R2 = 1GΩ and C = 1uF.
Since Vmax = 268mV = I(R1||R2).
So accordingly one can select values R1 = 20kΩ and I = 13.4u.
When Vg = 0 i.e. Vsg=2.5 and hence it is a stress and the Interface Trap voltage
increases with time and when Vg=2.5 i.e. Vsg=0, it is relax and hence the voltage
remains constant as shown is Figure 26. The interface traps don’t anneal during relax.
Figure 26: Interface Trap Voltage with Time and Gate Bias
39
4.4.2 Oxide Traps (Not) Generator
In response to gate voltage stress, the fixed trapped charge density may be
assumed to follow a double exponential response profile. Figure 27 shows the ideal
response of the oxide traps during stress and when the stress is removed.
Figure 27: Ideal Response of Oxide Traps During Stress and Relax
The above response can be captured by the transient equation
VNot(t) = Vpb[(1-exp(- 𝑡
𝜏1))u(t) – (1 – exp( -
𝑡−𝑇
𝜏2))u(t-T)] (53)
Where
VNot(t) = 𝑞𝑁𝑜𝑡(𝑡)
𝐶𝑜𝑥 (54)
The above equation can be generated in a circuit by a gate voltage-controlled switch
and an RC network shown in Figure 28.
40
Figure 28: Not Defect Generator Schematic
The maximum value of Not is taken to be 1.2x1012
.
Therefore VNot = 𝑞𝑁𝑜𝑡
𝐶𝑜𝑥 = 322mV.
During Stress, the switch is closed and the time constant τ1 = (R1||R2)C.
During Relax, the switch is open and the time constant τ2 = (R2)C.
Since Vmax = 322mV = I(R1||R2).
So accordingly one can select values R1 = 1MΩ, R2 = 5kΩ, C=1uF and I = 65u.
When Vg = 0 i.e. Vsg=2.5 and hence it is a stress and the Oxide Trap voltage
increases with time and when Vg=2.5 i.e. Vsg=0, it is relax period and hence the
voltage decreases as shown is Figure 29. The Oxide traps anneal when the stress is
removed.
41
Figure 29: Oxide Trap Voltage with Time and Gate Bias
4.5 Static NBTI
In static NBTI, the gate of the PMOS transistor is stressed for a long time and the
change in threshold voltage is observed with respect to time. The threshold voltage
increases monotonically with time as the defects keeps on increasing. The threshold
voltage shift shows two asymptotes: the slow response and the fast response [32]. The
inrerface traps (Dit) voltage is very slow and takes a lot of time to reach its maximum
value, whereas the oxide traps (Not) voltage reaches its saturation value way before
the Interface trap voltage. Using the model developed in this work, a similar pattern is
observed for the threshold voltage shift. Figure 30 shows the threshold voltage shift
generated using our model. The time constants used in the equations (51) and (53) are
chosen in such a way that the final value of oxide trap voltage saturates much before
the final value of the interface trap voltage.
42
The change in threshold voltage varies with time as a power-law [32].
∆𝑉𝑡ℎ ∝ 𝑡𝑛
The value of n varies depending of what type of defects dominates.
Figure 30: Static NBTI
4.6 Dynamic NBTI
In Dynamic NBTI, the gate of the PMOS transistor is allowed to switch from 0 to
VDD. When the source gate voltage is 0, the transistor is in strong inversion and
because of the gate stress oxide traps and interface traps are formed. When the stress
is removed i.e. if the source gate voltage is VDD, then the defects are annealed up to
certain extent because the trapped holes will be de-trapped as discussed in the earlier
sections. The interface traps are fixed but the oxide traps anneal when the stress is
removed. Therefore the threshold voltage degradation is annealed up to some extent.
43
Figure 31 shows the threshold voltage with time using the degradation model. The
fast component is because of the oxide traps whereas the slow component is due to the
interface traps [32].
Figure 31: Dynamic NBTI
44
CHAPTER 5
NBTI EFECTS IN CMOS CIRCUITS
5.1 CMOS Inverter
The CMOS inverter is a basic building block for digital circuit design. The input
to the inverter is connected to both a PMOS transistor and an NMOS transistor. When
the voltage of input A is low, the NMOS transistor is turned off and its channel is in a
high resistance state. The PMOS transistor is strongly on and its channel is in a low
resistance state and it allows the current to flow from the VDD to the output. It charges
the load capacitance. As the resistance between the supply voltage and output is low,
the voltage drop between the supply voltage and output due to a current drawn from
output is small resulting in a high voltage at the output [27].
Figure 32: CMOS Inverter
On the other hand, when the voltage of input A is high, the PMOS transistor is turned
off so it would not allow the current to flow from the VDD to the output, while the
NMOS transistor is completely turned on and, allowing the current to flow from
output to ground. The low resistance between output and ground keeps the voltage
45
drop due to a current drawn into output very small. This low drop results in a low
output voltage [27].
In short, inverter is a circuit where the output is inverted, such that when the input is
high (VDD), the output is low (0), and when the input is low, the output is high. The
NMOS and PMOS transistor goes through different regions as the input swings from
0 to VDD. The threshold voltage of both the transistors is important as the propagating
signal depends on the threshold voltage.
This CMOS inverter transfer characteristics is shown in Figure 33.
Figure 33: The CMOS Inverter Voltage Transfer Characteristics
There are various regions of operation when the input goes from 0 to VDD. When the
input is low, the M2 transistor is turned on while the M1 transistor is off. In the mid
values of input, both the transistors are turned on. When the input is high, the M2
transistor is off and M1 transistor is turned on. X and Y are the transition points. The
total intrinsic propagation delay of an inverter is given by td = (tpHL + tpLH)/2, where
tpHL and tpLH are explained in Figure 34.
46
Figure 34: Inverter Propagation Delay
The propagation delay depends on almost all the parameters like supply voltage,
threshold voltage, width, effective length, channel mobility etc.
5.1.1 NBTI Effects in CMOS Inverter
NBTI primarily occurs in p-channel MOSFET’s when a negative gate voltage
is applied and it is almost negligible when a positive voltage is applied. When the
source gate voltage is high, the transistor is in stress and when source gate voltage is
zero, the transistor is off. Although the defects increase with stress, but when the
stress is removed then interface traps are going to be annealed which recovers the
threshold voltage. This leads to an increase in the propagation delay. The NBTI can
also cause a spread in signals. The increase in spread of signals can cause serious
malfunction of the logic circuits and hence they will be failed [18].
The delay time td [53] is
47
td = 𝐶𝑉𝐷𝐷
𝐼 =
𝐶
1
2
𝑊
𝐿𝜇𝑒𝑓𝑓𝐶𝑜𝑥𝑉𝐷𝐷(1−
𝑉𝑡ℎ𝑉𝐷𝐷
)2 (55)
Therefore current decreases and delay time increases when threshold voltage
increases and mobility decreases.
In an inverter the NBTI causes a propagation delay and when a chain of inverters are
connected like in a ring oscillator circuit, the total propagation delay increases and
which causes the frequency of the ring oscillator to change.
Table 2 shows the phases of bias temperature instability in a CMOS inverter.
Table 2: Bias Stress Temperature in Inverter [53]
Vin Vout NMOS PMOS
0 VDD Relax(off) NBTI(stress)
VDD 0 PBTI(stress) Relax(off)
Figure 35 illustrates the various phases of NBTI in a CMOS inverter when the gate
voltage is swinging from 0 to VDD.
48
Figure 35: CMOS Inverter Degradation [53]
5.1.2 Inverter with the degradation model
The defect model developed in this model is connected to the gate terminals of
both PMOS and NMOS in the inverter. Although the PBTI effect in NMOS is very
less compared to that of NBTI in PMOS, the defect model for NMOS is also
connected.
49
When the gate voltage is 0V, the source gate voltage for PMOS is VDD hence it is
stress condition. When gate voltage is VDD, the PMOS is in off condition. Similarly
when the gate voltage is 0, the NMOS is off and when the gate voltage is VDD then the
NMOS is under stress. Although the PBTI effects are very minimum compared to
NBTI, both the effects are considered in this inverter simulation.
Figure 36: CMOS Inverter with and Without Degradation Models
Table 3: Component parameters of Inverter
Wp(nm) 12000
Wn(nm) 4000
L(nm) 1000
Inverter with and without the degradation model are shown in Figure 36. The
degradation model mimics the effects of NBTI in PMOS. The width and lengths of
PMOS and NMOS used in the inverter design are shown in table 3. When the stress is
applied, interface traps and oxide traps are formed which degrades the performance of
50
the PMOS by reducing its mobility and increasing absolute value of threshold voltage.
This causes a propagation delay in the inverter as shown in Figure 37. Although the
defects models are modeled for both PMOS and NMOS, the only major factor
contributing for propagation delay is the NBTI effect of PMOS transistors as the PBTI
in NMOS contributes very less effect. As the simulation results shows that the defects
cause an asymmetry in the rise and fall times and the propagation time for the signal
increases causing a delay. So when the inverters are connected in a chained fashion
like in a ring oscillator, the total propagation delay increases and frequency of the ring
oscillator decreases.
Figure 37: CMOS Inverter Output with and Without Stress
5.2 Ring Oscillator
A ring oscillator is a device composed of an odd number of Inverters connected in
a positive feedback loop, whose output oscillates between two voltage levels 0 and
VDD. A block diagram of an n-stage inverter ring oscillator is shown in Figure 38.
51
Figure 38: An n-Stage Ring Oscillator
The inverters are connected in a chain fashion in stages; the inverters are basically
connected in cyclic order, so that the output of the last inverter is fed back as an input
to the first inverter of the chain. As there will be an inversion for every stage of
inverters, the output will also be inverted if the number of stages is an odd number.
There is a certain time delay for every inverter for the signal to pass from input to
output. It is also the same time to charge and discharge the capacitance at the output
node. Thus every inverter in the chain has a time delay for its output to show up. As
the number of inverters increase, the gate delay increases which decreases the
frequency of the ring oscillator. The main reason for the oscillation is the odd number
of inverters and output of last inverter feeding back into the input. A real ring
oscillator is almost a self-start circuit which only requires power to operate; above a
certain threshold voltage, oscillations begin spontaneously. To increase the frequency
of oscillation, ring oscillator should be made from a smaller number of inverters.
The frequency of oscillation at each stage is inversely proportional to the number of
stages in the ring oscillator circuit and the gate delay of each individual gate. As the
number of stages increase, the frequency of oscillation decreases.
52
fosc = 1
2𝑛𝑡𝑑 (56)
Where “n” denotes the number of inverters used, td is the time delay of each inverter.
The frequency of the ring oscillator depends on process parameters like the width,
length, threshold voltage etc. The propagation delay of an inverter circuit can be
obtained by measuring the time period of the oscillator.
With the gate stress and at elevated temperatures, the interface traps and oxide traps
are formed which degrades the performance of the transistor by increases the
threshold voltage and degrading other important parameters. As the threshold voltage
increases, the value of (VDD–Vth) decreases and hence it increases the delay according
to the equation (55).
In this work, a ring oscillator circuit of 5 inverters chain is designed as shown in
Figure 39 and is simulated with and without the stress. In spectre simulations, an
initial condition is given to the output node and then the output oscillates
automatically. The results are also compared with the results of only Vt mode. The
frequency of the ring oscillator without stress is 246.5MHz and we observed that the
frequency with stress as decreased. The frequency of the ring oscillator with stress
i.e., when the defects model is applied is 204MHz. The frequency when just the Vt
model is applied is 198.5MHz.
53
Figure 39: Ring Oscillator Schematic with 5 Inverters
Figure 40: Ring Oscillator Output Waveform with and Without Stress
The transient response with and without stress are shown in Figure 40. In the Vt
model, the gate bias dependency in the subtreshold region is not considered. The
frequency response is compared in Figure 41 and Figure 42 respectively.
54
Figure 41: Frequency Response without Stress
Figure 42(a): Frequency Response with Stress (Defects Model)
Figure 42(b): Frequency Response with Stress (Vt Model)
55
The model is also applied to test the single event transient (SET) pulse broadening
and we see the pulse broadening with stress as shown in Figure 45.
Figure 43: Modeling SET Pulse Broadening
Figure 45(a): SET Pulse Broadening Without Stress
Figure 45(b): SET Pulse Broadening with Stress
No Stress
Stress
56
CHAPTER 6
CONCLUSION
Negative bias temperature instability (NBTI) is a serious problem in PMOS
transistors. The defects caused by the gate stress and elevated temperatures should be
considered in the circuit simulations. But the defects change with time and they are
not constant, so it is very hard for the designer to include the defects every time. So in
this work a model is developed to which the inputs are the maximum value of the
density of interface traps and oxide traps. The model generates a defect potential
which is exactly the change is threshold voltage caused by the defects. The defects are
gate voltage and time dependent.
The MOSFET Sub-threshold current method is used to evaluate the interface
traps and oxide traps in PMOS transistor. The degradation model in modeled in
cadence and validated the same with the simulations in Silvaco. The current-voltage
curves in both cadence and Silvaco are compared and both the results fit each other. A
shift is observed in the threshold voltage when the degradation model is introduced
which is obvious as the shift is caused by the interface traps and the oxide traps. The
simulation results produced by the model discussed in this work are compared to
various other results by other authors. The same methodology is replicated for NMOS
transistor and a degradation model is developed for NMOS transistor. After the
device level simulations, the degradation models were introduced at the gate terminal
of PMOS and NMOS in an inverter. The threshold voltage degradation is very less for
a NMOS transistor. A ring oscillator circuit of 5 inverters chain is designed and the
frequency of the ring oscillator is observed with and without stress. The frequency of
57
ring oscillator is decreased by applying stress which is because of the increase in
threshold voltage and hence increase in gate delay. The Vt only model result is also
compared with that of the defects model. There is a change is frequency in both the
models. The Vt only model doesn’t include the gate voltage dependency in the
subtreshold region. The model is also used to see the single event transient pulse
broadening and it gave a satisfactory result as observed by some authors.
This model captures transistor response using fixed trapped charge (Not) and interface
traps densities (Dit) as the input parameters. The model is external to the transistor,
thus easily implemented with time and bias dependence but without modifications to
compact model. This is very advantageous to the designers.
58
REFERENCES
[1] H.J. Barnaby,“Total-Ionizing-Dose Effects in Modern CMOS Technologies”
IEEE Trans. Nucl. Sci. Vol. 53, No. 6, December 2006.
[2] Boualem Djezzar, Hakim Tahi, and Arezki Mokrani, “Why Is Oxide-Trap
Charge-Pumping Method Appropriate for Radiation-Induced Trap Depiction in
MOSFET?” IEEE Trans. on Device and Materials Reliability, Vol. 9, No. 2, June
2009
[3] V. Huard, M. Denais, and C. Parthasarathy, ``NBTI Degradation: From
Physical Mechanisms to Modelling,'' Microelectron.Reliab., vol. 46, no. 1, pp. 1-
23, 2006.
[4] P. M. Lenahan and P. V. Dressendorfer, “Hole traps and trivalent silicon
centers in metal/oxide/silicon devices,” J. Appl. Phys., vol. 55, pp. 3495, 1984.
[5] Y. F. Chen, M. H. Lin, C. H. Chou, W. C. Chang, S. C. Huang, Y. J. Chang,
K. Y. Fu, M. T. Lee, C. H. Liu, and S. K. Fan, “Negative bias temperature
instability (NBTI) in deep sub-micron P+-gate pMOSFETs,” in Proc. IRW, Lake
Tahoe, CA, pp. 98–101, Oct. 2000.
[6] C. H. Liu, M. T. Lee, C. Y. Lin, J. Chen, Y. T. Loh, F. T. Liou, etc,
“Mechanism of threshold voltage shift (∆Vth) caused by negative bias
temperature instability (NBTI) in deep submicrometer pMOSFETs,” Jpn. J. Appl.
Phys., vol. 41, pp. 2423-2425, April 2002.
[7] M. A. Alam, H. Kufluoglu, D. Varghese, and S. Mahapatra, “A
comprehensive model for PMOS NBTI degradation: Recent progress,”
Microelectron. Reliab., vol. 47, no. 6, pp. 853–862, Jun. 2007.
[8] K. O. Jeppson and C. M. Svensson, “Negative bias stress of MOS devices at
high electric fields and degradation of MOS devices,” J. Appl. Phys., vol. 48, pp.
2004–2016, 1977.
[9] S. Mahapatra, P. B. Kumar, and M. A. Alam, “A new observation of
enhanced bias temperature instability in thin gate oxide p-MOSFETs,” in IEDM
Tech. Dig. pp. 196–202, 2003.
[10] Dieter K. Schroder, “Negative bias temperature instability: What do we
understand?” Microelectronics Reliability, vol. 47, pp. 841-852, 2007.
[11] M. A. Alam, S. Mahapatra, “A comprehensive model of PMOS NBTI
degradation,” Microelectronics Reliability, vol. 45, pp. 71-81, 2005.
59
[12] “MOSFET.” Wikipedia. Retrieved October 15, 2014, from
http://en.wikipedia.org/wiki/MOSFET.
[13] A. S. Sedra, K. C. Smith, “Microelectronics circuits”, Oxford University
Press, 2004.
[14] A. Narr, and A. Lill, “Lifetime Prediction for PMOS and NMOS Devices
Biased on A Degradation Model for Gate-Bias-Stress," Microelectronics
Reliability, vol. 37, pp. 1433-1436, 1997.
[15] Y. H. Lee, S. Jacobs, S. Stadler, N. Mielke, and R. Nachman, “The impact of
PMOST bias-temperature degradation on logic circuit reliability for
performance," Microelectronics Reliability, vol. 45, pp. 107-114, 2005.
[16] A. T. Krishnan, V. Reddy, and S. Krishnan, “Impact of charging damage on
negative bias temperature instability,” in IEDM Tech. Dig., pp. 39.3.1–39.3.4.,
2001.
[17] Harris, H.R.; Choi, R.; Lee, B.H.; Young, C.D.; Sim, J.H.; Mathews, K.;
Zeitzoff, P.; Majhi, P.; Bersuker, G.; “Recovery of NBTI degradation in
HfSiON/metal gate transistors, Integrated Reliability Workshop Final Report,”
2004 IEEE International, pp. 132 – 135, Oct. 2004 .
[18] Dieter K. Schroder, “Negative bias temperature instability: Road to cross in
deep submicron silicon semiconductor manufacturing,” J. Appl. Phys., vol. 94, 1
July 2003.
[19] S. Zafar, “Statistical mechanics based model for negative bias temperature
instability induced degradation," Journal of Applied Physics, vol. 97, pp. 1-9,
2005.
[20] Robert Entner, “Modeling and Simulation of Negative Bias Temperature
Instability,” PhD Dissertation, Institute of Microelectronics, 2007.
[21] Chen G., Li M.F., Ang C.H., Zheng J.Z., Kwong D.L., “Dynamic NBTI of
p-MOS transistors and its impact on MOSFET scaling,” IEEE Electron Dev Lett,
vol. 23, pp. 734–6, 2002.
[22] Alam M.A., “A critical examination of the mechanics of dynamic NBTI for
pMOSFETs,” IEDM Tech Dig, pp. 346–9, 2003.
[24] Yang, T. Shen, C. Li, M.F. Ang, C.H. Zhu, C.X. Yeo, Y.-C. Samudra, G.
Kwong, D.-L., “Interface trap passivation effect in NBTI measurement for p-
MOSFET with SiON gate dielectric, Electron Device Letters,” IEEE, Vol. 26, pp.
758- 760, Oct.2005.
60
[26] Xiaojun Li, Jin Qin, Bing Huang, Xiaohu Zhang, and Joseph B. Bernstein,
“A New SPICE Reliability Simulation Method for Deep Submicrometer CMOS
VLSI Circuits,” IEEE TRANSACTIONS ON DEVICE AND MATERIALS
RELIABILITY, vol. 6, No. 2, pp. 247-257, June. 2006.
[27] “CMOS.” Wikipedia. Retrieved September 30, 2014, from
http://en.wikipedia.org/wiki/CMOS.
[32] A. Neugroschel, G. Bersuker, R. Choi, C. Cochrane, P. Lenahan, D. Heh, C.
Young, C.Y. Kang, B.H. Lee, and R. Jammy, “An Accurate Lifetime Analysis
Methodology Incorporating Governing NBTI Mechanisms in High-k/SiO2 Gate
Stacks” in Proc. Int. Electron Devices Meeting, San Francisco, CA, Dec. 2006,
pp. 317-320.
[33] C. Leroux, J. Mitard, G. Ghibaudo, X. Garros, G. Reimbold, B. Guillaumor,
F. Martin, “Characterization and modeling of hysteresis phenomena in high-k
dielectrics,” IEEE Intl Electron Devices Meeting Tech Digest. pp. 737-740, 13-15
Dec. 2004.
[34] S. Zafar, A. Callegari, E. Gusev, M.V. Fischetti, “Charge trapping in high-k
gate dielectric stacks,” IEEE IEDM, pp. 517-520, 2002.
[35] G. RibePb, M.Miillerd, S. Bruykre, D. Roy ,M. Denaisa, V. Huardd,
TSkotnicki, G. Ghibaudob, “Characterization of Vt instability in hafnium based
dielectrics by pulse gate voltage techniques,” IEEE, vol. 5, pp. 89-92, 2004.
[36] Gurfinkel, M.; Suehle, J.; Bernstein, J.B.; Shapira, Y.; Lelis, A.J.; Habersat,
D.; Goldsman, N.; “Ultra-Fast Characterization of Transient Gate Oxide Trapping
in SiC Mosfets,” IEEE International pp. 462 – 466, April 2007.
[38] C.D. Young, Y. Zhao, M. Pendley, B.H. Lee, K. Matthews, J.H. Sim, R.
Choi, G.A. Brown, R.W. Murto and G. Bersuker, “Ultra-Short Pulse Current-
Voltage Characterization of the Intrinsic Characteristics of High k Devices”, Jpn.
J. of Appl. Phys., vol. 44, No. 4B, 2437, 2005.
[39] KERBER A.; CARTIER E.; PANTISANO L.; DEGRAEVE R.;
GROESENEKEN G.; MAES H. E., SCHWALKE U. , “Charge trapping in
SiO2/HfO2 gate dielectrics: Comparison between charge-pumping and pulsed
ID–VG,” Microelectronic Engineering, vol.72, pp. 267-272, April 2004.
[40] C. D. Young, R. Choi, J. H. Sim, B. H. Lee, P. Zeitzoff, Y. Zhao, K.
Matthews, G. A. Brown, and G. Bersuker, “Interfacial layer dependence of
HFSI/sub x/O/sub y/ gate stacks on V/sub T/ instability and charge trapping using
ultra-short pulse in characterization,” presented at the 43rd Annual IEEE Intl.
Reliability Physics Symp. Proc, pp.75-79, 2005.
61
[41] Kang, Kunhyuk Kim, Keejong Islam, Ahmad E. Alam, Muhammad A. Roy,
Kaushik, “Characterization and Estimation of Circuit Reliability Degradation
under NBTI using On-Line IDDQ Measurement,” Design Automation
Conference, 2007. DAC '07. 44th ACM/IEEE, Publication, pp. 358 – 363, June
2007.
[42] L. Tsetseris, X. Zhou, D. Fleetwood, R. Schrimpf, and S. Pantelides,
``Physical Mechanisms of Negative-Bias Temperature Instability,''
Appl.Phys.Lett., vol. 86, no. 14, pp. 1-3, 2005.
[43] T. Grasser, W. Gös, V. Sverdlov, and B. Kaczer, ``The Universality of NBTI
Relaxation and its Implications for Modeling and Characterization,'' in Proc.
Intl.Rel.Phys.Symp., pp. 1-13, 2007.
[44] D. K. Schroder, Semiconductor Material and Device Characterization, 3rd
Edition. Wiley Interscience, 2006.
[45] R. A. B. Devine, J.-L. Autran, W. L. Warren, K. L. Vanheusdan, and J.-C.
Rostaing, ``Interfacial Hardness Enhancement in Deuterium Annealed 0.25 mu m
Channel Metal Oxide Semiconductor Transistors,'' Appl.Phys.Lett., vol. 70, no.
22, pp. 2999-3001, 1997.
[46] A. H. Edwards, ``Interaction of H and H2 with the Silicon Dangling Orbital
at the <111> Si/SiO2 Interface,'' Phys.Rev.B, vol. 44, no. 4, pp. 1832-1838, 1991.
[47] V. Reddy, A. T. Krishnan, A. Marshall, J. Rodriguez, J. , S. Natarajan, T.
Rost, S. Krishnan, “Impact of Negative Bias Temperature Instability on Digital
Circuit Reliability,” IEEE Reliability Physics Symposium Proceedings, 2002.
40th Annual, pp. 248-254, 2002.
[48] J. Stathis and S. Zafar, ``The Negative Bias Temperature Instability in MOS
Devices: A Review,'' Microelectron.Reliab., vol. 46, no. 2-4, pp. 270-286, 2006.
[49] Y. Miura and Y. Matukura, ``Investigation of Silicon-Silicon Dioxide
Interface Using MOS Structure,'' Jpn.J.Appl.Phys., vol. 5, p. 180, 1966.
[50] A. Goetzberger and H. Nigh, ``Surface Charge After Annealing of Al- SiO2 -
Si Structures Under Bias,'' Proc.IEEE, vol. 54, no. 10, p. 1454, 1966.
[51] E. Deal, M. Sklar, A. S. Grove, and E. H. Snow, ``Characteristics of the
Surface-State Charge (Qss) of Thermally Oxidized Silicon,'' J.Electrochem.Soc.,
vol. 114, no. 3, p. 266, 1967.
[52] Y. Mitani, M. Nagamine, H. Satake, and A. Toriumi, ``NBTI Mechanism in
Ultra-Thin Gate Dielectric - Nitrogen-Originated Mechanism in SiON,'' in Proc.
Intl. Electron Devices Meeting, pp. 509-512, 2002.
62
[53] D. K. Schroder, “Advanced MOS Devices”, class notes, Arizona State
University, Tempe, 2011.
[54] C. R. Helms and E. H. Poindexter, ``The Silicon-Silicon Dioxide System: Its
Microstructure and Imperfections,'' Reports on Progress in Physics, vol. 57, no. 8,
pp. 791-852, 1994.
[55] J. P. Campbell, P. M. Lenahan, A. T. Krishnan, and S. Krishnan,
``Observations of NBTI-Induced Atomic Scale Defects,'' IEEE
Trans.Dev.Mat.Rel., vol. 6, no. 2, pp. 117-122, 2006.
[56] J. Campbell, P. Lenahan, A. Krishnan, and S. Krishnan, ``Direct Observation
of the Structure of Defect Centers Involved in the Negative Bias Temperature
Instability,'' Appl.Phys.Lett., vol. 87, no. 20, pp. 1-3, 2005.
[57] P. Lenahan, T. Mishima, J. Jumper, T. Fogarty, and R. Wilkins, ``Direct
Experimental Evidence for Atomic Scale Structural Changes Involved in the
Interface-Trap Transformation Process,'' IEEE Trans.Nuclear Science, vol. 48,
no. 6, pp. 2131-2135, 2001.
[58] L. A. Ragnarsson and P. Lundgren, ``Electrical Characterization of Pb
Centers in (100) Si/SiO2 Structures: The Influence of Surface Potential on
Passivation During Post Metallization Anneal,'' J.Appl.Phys. vol. 88, no. 2, pp.
938-942, 2000.
[59] V. Reddy, J. Carulli, A. Krishnan, W. Bosch, and B. Burgess, “Impact of
Negative Bias Temperature Instability on Production Parameter Drift,” IEEE ITC
International Test Conference, pp. 148-155, 2004.
64
go atlas
#set Niti=2.3e10
# D is Niti/.023 where .023 eV = phiB/20 and PhiB=.46 for the nmos2 str
set Not=1.2e12
mesh inf=pmos_tsmc.str
model print SURFMOB
material region=1 eg300=1.12 nc300=3.2e19 nv300=2e19 mun=455.3
mup=158.7 permittivity=11.7 affinity=4.05 taun0=1 taup0=1
contact name=source neutral
contact name=drain neutral
contact name=gate workfun=5.17
interface qf=$Not
method 2nd tol.time=1e-6
output con.band val.band band.params charge
solve init
save outf=pmos_0V_$"Not"_$"Niti".str
# save structure for off state Vds=50mV
solve vdrain=-0.05 name=drain
save outf=pmos_sd_p05V_$"Not"_$"Niti".str
solve vgate=0
# save IV for dc Id-Vgs linear response
log outf=T_PMOS_1pe12_1e12.log
solve vgate=0 vstep=-0.1 vfinal=-2.5 name=gate
# save structure for strong-inv, linear Vds=50mV
save outf=pmos_tsmc_2p5.str
quit